Evaluating the 2023 QBs: The case for Will Levis

I want to be clear upfront that in five years, when people look back on my quarterback predictions for the 2023 draft class, my official opinion is that I don’t care for Will Levis and that I do not think he will work out. The reason I want to be clear about that is because I’ve softened a bit on that stance, and this preview is going to be extremely wishy-washy on the subject, with good reason. But I like to have some accountability for my predictions without “both-sidesing” things too much, so officially, I think he’s bad.

Let’s start with the reasoning behind my official opinion. I don’t like the Levis archetype, which is quite common in college football, and which I refer to as “Big Arm, Slow Process.” Levis suffers from an issue common to those capable of putting on the laser show, wherein your arm allows you to get by with late throws. You don’t need to be as good on your reads because once you do find a guy, you can just rifle it in there, and this will work just fine at most levels of football.

The BASPs are easily identifiable as they share several traits in common. The big arm is a prerequisite, but you will also tend to see an uptick in interceptions, and here we refer to the QBOPS+ column on the draft sheet. The thing to look for is the difference between regular old QBOPS (an excellent 1.023 for Levis) and the + version (a rather terrible .846), which adjusts for interception percentage. You’ll note that Levis is color coded red, as the -.177 difference between the two numbers is the largest in the class. Levis threw far too many picks.

The BASP profile does come with some positives, but those positives tend not to translate as well to the next level. When I grade quarterbacks, I like to see value first come from a high completion percentage, but that’s not the only way to generate value, and the BASPs tend to be big bombers with high Yards per Completion numbers. Levis’ .637 QBSLG isn’t as massive as we sometimes see, but it’s big enough, well above my standard .600 threshold. That big arm is able to attack parts of the secondary that are off-limits to most college quarterbacks. Safeties aren’t as athletic, there’s a ton of space because of how the hashes work, and as a result, bombs are a fertile feeding ground for the BASP crew.

Unfortunately, what works wonders in college doesn’t always translate to the NFL. The field dimensions and the superior nature of NFL athletes in the secondary mean that you can’t be late on your throws regardless of your arm talent, or you will pay for it. There is better coverage in the deep secondary, where truly great athletes are just waiting for you to try to lock on to a deep threat and throw up a prayer. The picks will tick up even more, and the big plays will start to dissipate. There are some additional issues with Levis which we will get to in a moment, but the BASP profile is the reason I approached him predisposed not to like him.

Jordan Love, by the way, was a prototypical BASP as a prospect. He possesses an absolute rocket for an arm, but also a long delivery, a lack of accuracy, and an incredible penchant for interceptions, even in his good season. Love was absolutely dynamic hitting guys downfield for huge gains, (his 2018 QBOPS slash line of .378/.655/1.032 slash line is quite similar to Levis’) but that lack of short-throw accuracy still bothers me to this day. I hope that Love has made progress sitting behind Rodgers, and that the Packers have run a smart development program behind the scenes and harnessed that arm. But I’m still skeptical, and when I first saw Levis, I saw Love.

Then, a funny thing happened.

There is a well-known cognitive bias called “anchoring” which impacts everyone on earth, and makes it extremely difficult to change your mind on any topic. Anchoring doesn’t just lock you into whatever viewpoint you’ve decided on, it also sets that viewpoint as the baseline for future decisions. If you are attempting to guess how many people there are in an arena, and your underlying assumption is that there are 12, and you receive evidence that no, there are more than 12, you are much more likely to update your guess to something like a still low 100 and not 1000, or 10,000.

One of the few methods for de-anchoring is actually to make your arguments out loud, not written down, not on the internet. First, you have to realize you have a built-in bias in the first place, but once you do, saying the arguments out loud is quite helpful, because it’s difficult to articulate things you don’t really believe out loud. I ran into this while recording a podcast on the quarterbacks of this class, and as I got into Levis, I quickly realized that my case against him was not nearly as strong as I had assumed, and not supported by all of the metrics I had assembled. I ended up re-recording the entire segment because upon re-listening, and looking at my spreadsheets, I was being unfair, and I have spreadsheets in the first place specifically so that I am NOT unfair.

The Pro-Levis Argument

And so let us dive into the strong case in favor of Levis, which is MUCH more complicated than I ever imagined it could be. And let’s start with accuracy, because Levis WAS accurate. He didn’t quite clear my .400/.600 threshold, but his .386/.637/1.023 QBOPS line is quite good, and he ranked 29^th out of 114 qualifiers in completion percentage in 2022. Moreover, only seven of the quarterbacks who were more accurate also had a higher yards per completion average. My first instinct was to dismiss his close-but-not-quite accuracy as a result of an easy schedule, but of course, Kentucky, and SEC school, doesn’t play an easy schedule.

While their .197 SOD (Note: For more on SOD and WRGPC, please refer to the glossary) isn’t in the same league as C.J. Stroud or Hendon Hooker, Levis actually played a tougher schedule than Bryce Young (and had a higher completion percentage), and in fact played a tougher schedule than the vast majority of quarterbacks in the class, ranking 25^th. And while several other quarterbacks were bolstered by their receivers in 2022, Levis was not. His Wide Receiver Grade per Completion was just 68, only two points better than average, and more importantly, those receivers dropped 7.5% of their targets, which is quite a lot. (For context, Clayton Tune’s receivers were the worst among draft-eligible QBs with an 8.5% rate, while Jaren Hall’s were the best at just 4.3%). Levis was, in a vacuum, more accurate than his raw numbers.

Going back to Jordan Love for a moment, his penultimate season was much better than his final season, where his supporting cast all left and he threw the most picks in all of college football, and I’m also always skeptical of quarterbacks who decline at the end of their college careers, even if there are some mitigating factors. Even adjusting for the quality of defense faced, and a decline in receiving talent, Love still regressed from a good, if not great 108 QwOBA+ in 2018 to a very poor 94 in his final season.

There is, with Levis, a notion that something similar happened: that his second-to-last season was superior and he suffered a decline in 2022. As it turns out, that is absolutely not the case, and his final season was much better than anyone understands. In 2021 Levis was also just shy of the .400/.600 line, posting a .389/.594/.984 QBOPS line that was marginally more accurate, but lacked the dynamism of the 2022 season and actually featured more interceptions as a percentage of total targets. I think people see Levis’ gaudier touchdown total from 2021 (24, vs. only 19 in 2022) and slightly higher completion percentage (66% vs. 65.4% in 2022), while also not realizing that in terms of interception percentage, he actually improved slightly in 2022. On the surface, that narrative makes some sense.

Dig a bit deeper though, and the story changes. First, his strength of defense faced rose drastically from 2021 (.022, an almost exactly average schedule) to 2022 (.197). That’s an enormous increase in difficulty. Second, and more importantly, the quality of the receivers Levis was targeting changed dramatically for one specific reason: Wan’Dale Robinson. Robinson was one of PFF’s highest graded receivers in 2021, and in this case, it’s hard to argue with that assessment. He put up a great season, with 104 catches for 1334 yards and a stellar catch percentage. He was eventually selected in the early 2^nd round of the 2022 draft by the Giants, and after getting off to a slow start due to an injury, he really started to make his presence felt starting in Week 6. Unfortunately, he would tear his ACL against the Lions in Week 10 after recording a nine-catch, 100-yard performance.

Because Robinson was so good, graded so highly, and targeted so much, Levis is actually adjusted down fairly heavily by QwOBA+ for that year. Levis’ production also reflects Robinson’s chief skillset as a dynamic slot receiver who thrives as more of a possession player. In 2022, Kentucky’s receivers were objectively worse and graded worse, but Levis, to his credit, did a much better job of spreading the ball around and attacking more downfield. His Wide Receiver Grade per Catch went from 74.38 in 2021 to just 68.06 in 2022 (66 is average).

In short, Levis not only got better in 2022, he got significantly better. He played against much better defenses while targeting inferior receivers and boosted his QwOBA+ from 97 to 112. My Love comparison fell completely apart as I was saying it out loud into a microphone.

There’s one final piece of pro-Levis evidence worth mentioning as well, which is the S-2 Cognition Test. (If you are not familiar with the S-2, please see this post in The Athletic, and my Bryce Young write-up.) The S-2 has become a major talking point this draft cycle, and while I am not completely sold on its usefulness, I think it’s worth noting that Will Levis scored quite well on it.

“Processing” remains a difficult attribute to scout. I often use interception percentage as a proxy, but it’s certainly an imperfect one, and picks can be caused by several factors. None of those factors are good, exactly, but some are more easily fixable than others. Brett Favre was a great processor who had too much confidence in his arm. Perhaps Levis is more along those lines.

What to believe on Levis?

And so I’m not entirely sure what to believe on Will Levis, but at the very least, I now understand the intrigue and the pro-Levis case. Honestly, it’s a compelling one. He’s not quite the physical specimen that Anthony Richardson is, but if someone wanted to compare him to a slightly more pick-prone Justin Herbert, I don’t think it would be the worst comp.

There are two other big negatives I should mention because, as I said earlier, I’m still generally down on him. The first is his pressure-to-sack percentage, which is an absolutely terrible 26.8%, second-worst in the class to Hendon Hooker. It’s a stat that tends to translate to the NFL, with Justin Fields as the prime example of someone who struggled and still struggles in that area. Teams with poor offensive lines: beware.

The other is that while Levis did face a fairly tough schedule, he also fattened up on the dregs of that schedule, with good games against Miami (OH), Northern Illinois, and Ole Miss. He struggled mightily against the big boys in Georgia, Florida, and Tennessee, and had a disturbingly bad game against Youngstown State where he threw two picks. To the extent he has a signature win on his resume, it’s likely his game against at Missouri, but even there, he only completed 13 passes for 170 yards (though with three touchdowns).

More than anything, it would likely help Levis to land in a spot with some competent quarterback development and not totally bereft of talent. Justin Fields winding up on the Bears was a tragedy for him and exacerbated all of his weaknesses, and there are a few teams that would, I suspect, hurt Levis immensely (including Indianapolis, which seems somewhat likely). But he’s a fun player with plenty of upside, and I can’t help myself from wondering what might happen if he falls to 15.